1. Field of the Invention
This invention relates to a device and method for detecting ionizing radiation, and more particularly to a semiconductor radiation detector with enhanced charge collection for reducing low-energy tailing effects.
2. Description of Related Art
High-resistivity semiconductor radiation detectors are widely used for detecting ionizing radiation due to their ability to operate at room temperature, their small size and durability, and other features inherent in semiconductor devices. Such detectors are used in a wide variety of applications, including medical diagnostic imaging, nuclear waste monitoring, industrial process monitoring, and space astronomy. Ionizing radiation includes both particulate radiation, such as alpha or beta particles, and electromagnetic radiation, such as gamma or x rays.
Conventional semiconductor radiation detectors are generally referred to as "planar" detectors. As shown in FIG. 1, the architecture of such planar detectors 100 typically consists of a slab of semiconductor crystal 102 with metal covering two opposing surfaces of the slab to form two electrodes, a cathode 104 and an anode 106. In one configuration, the anode 106 is connected to external signal conditioning circuitry 108 and to ground 110, and the cathode 104 is connected to an external voltage source 111. A bias voltage across the electrodes 104, 106 creates an internal electric field. Electron and hole "charge clouds" generated within the semiconductor crystal 102 by an ionizing radiation 112 absorbed within the slab of semiconductor crystal 102 are swept toward the anode 106 and cathode 104 electrodes, respectively. These moving electron and hole clouds create charge-pulse signals in the external signal conditioning circuitry 108.
If all the electrons and holes generated by the ionizing radiation 112 reach their respective electrodes (i.e., the electrons reach the anode 106 and the holes reach the cathode 104), the output charge signal will exactly equal the charge from the energy deposited within the crystal 102. Because the deposited charge is directly proportional to the energy of the ionizing radiation 112, the semiconductor radiation detector 100 provides a means for measuring the energy of the ionizing radiation 112. The ability to measure this energy is an important function of radiation detectors.
Planar radiation detectors, however, suffer from a serious drawback: because of limitation in the transport properties of the bulk semiconductor crystal 102, some of the electrons and holes are generally lost by being trapped as they sweep toward their respective electrodes. Thus, the amplitude of the output charge signal becomes dependent on the position within the crystal at which the ionizing radiation is absorbed. Generally, the amplitude is less than the charge deposited by the ionizing radiation 112, resulting in a corresponding reduction of energy measurement accuracy as well as poor resolution and reduced peak efficiency. This loss (or trapping) of charge in a radiation detector results in asymmetrical spectral peak shapes known as "low-energy tailing."
As stated above, in a semiconductor radiation detector, when an ionizing event occurs, electrons are swept toward the anode 106 and holes toward the cathode 104. In a typical experimental arrangement, with the cathode 104 facing the source of the radiation, many ionization events occur over some accumulation period, and the resulting charge signal pulses are detected and then displayed in a histogram. In an ideal detector, in which there is no low-energy tailing, all the pulses would be directly proportional to the energy of the ionizing radiation 112. This would result in a histogram like that of FIG. 2, in which counts per channel are plotted versus charge signal pulse amplitude. As can be seen in FIG. 2, the energy histogram exhibits no tailing, because the energy peak (or "photopeak") 202 appears as a straight vertical line at a single energy level, E, equal to the energy level of the ionizing radiation 112. Thus, all the charge signal pulses have an amplitude equal to the energy level E of the ionizing radiation 112, and no charge is lost in any single pulse.
Curves A and B of FIG. 3 illustrate two idealized cases of low-energy tailing in a non-ideal detector. Curve A represents the histogram distribution that would result if the ionizing radiation were absorbed uniformly throughout the crystal, as would occur with a very low absorption coefficient of the crystal. Curve B represents the more typical situation, where absorption is large near the cathode and drops off exponentially as the ionizing event moves in a direction away from the cathode within the crystal. In both Curves A and B, there is a maximum signal 302 corresponding to full charge collection (at amplitude "E") and pronounced low-energy "tails" 304, 306.
FIG. 4 shows an energy histogram exhibiting pronounced low-energy tailing for an actual semiconductor radiation detector made from Cadmium-Zinc-Telluride (CdZnTe) irradiated with gamma rays from a cobalt-57 (".sup.57 Co") radiation source. This detector had area dimensions of 6.1 mm by 6.1 mm and a thickness of3 mm. Its bias voltage was -500 volts. The data values in FIG. 4 are spread-out by electronic noise, an effect that was not considered in plotting the idealized curves of FIG. 3. As with Curves A and B of FIG. 3, the histogram of FIG. 4 has a pronounced low-energy tail 404.
Because of the deleterious effects of low-energy tailing in semiconductor detectors, much effort has gone into attempting to solve this problem. One approach to reducing the tailing effect in semiconductor detectors is to reduce the dependence of the signal pulse-charge amplitude on the position at which the ionizing radiation is absorbed. This can be accomplished, in principle, by contriving to limit to a small distance the region in which charge is induced on one electrode by a charge cloud in front of that electrode. If this is accomplished, a charge cloud generated by an ionizing event induces little charge on the electrode until it becomes very near the electrode, after which the charge cloud induces essentially all of its charge on that electrode. This approach is especially useful for semiconductors in which the transport properties of one carrier type (e.g., electrons) are much better than those of the other type (holes in this example). These transport properties are expressed by a "mobility-lifetime product." The ratio of the transport properties of one type carrier (e.g., holes) to those of the other type carrier (e.g., electrons) is expressed as the "mobility-lifetime-product ratio." Thus, the general approach described above is useful for all mobility-lifetime-product ratios, but is most useful for semiconductors having a large ratio of the larger mobility-lifetime product divided by the smaller. Semiconductors for which the mobility-lifetime-product ratio is greater than 10 include cadmium-zinc-telluride, cadmium-telluride, and mercury-iodide.
An early effort aimed at minimizing low-energy tailing using the above approach employed a semiconductor detector having a hemispherical configuration. See, e.g, H. L. Malm, et al., "Gamma-Ray Spectroscopy with Single-Carrier Collection in High Resistivity Semiconductors, " Appl. Phys. Lett., vol. 26, at 344-46 (1975). In Malm's detector, a large hemispherical surface of the cadmium-telluride was metallized to form the cathode. The anode formed a small circle at the center of the flat cross-section of the hemisphere. A bias voltage applied across these electrodes produced an internal electric field that varied from a low value near the cathode to a high value near the small anode. The electric field lines were thus concentrated near the central point by the spherical geometry. A result of this electric field concentration is that electrons move much faster in the close vicinity of the anode than in the remainder of the detector. Because the charge induced on the anode is inversely proportional to the square of the distance from the charge cloud to the anode, most of the charge is induced when the charge cloud is in the vicinity of the anode. The signal charge pulse amplitude is thus relatively insensitive to the position of the ionizing event in the detector.
The Malm approach, however, has several disadvantages. First, while energy resolution is improved over the planar detector, significant tailing remains. Second, it is difficult and thus costly to fabricate a semiconductor crystal having a hemispherical configuration. Third, the configuration cannot be applied to monolithic detector array structures. In consequence, this detector has achieved little, if any, commercial acceptance.
A second approach achieves a reduction in tailing by using a planar structure in which the anode is in two sections, a 0.5 mm diameter circle and a ring surrounding and at the same voltage as the circle, and the cathode covers the opposite surface. See, e.g., F. P. Doty, et al., "Pixilated CdZnTe Detector Arrays," Nucl. Instruments & Methods in Physics Research, vol. A 353, at 356-60 (1994). The charge induced by an electron cloud is shared between the small circle and the ring, such that the charge induced on the circle is very small until the charge comes very close to the circle. The full charge is then induced on the circle within a distance comparable to the pixel dimensions.
This second approach also suffers from a significant disadvantage, in that it results in a very low collection efficiency. This result stems from the fact that only charge clouds directly above the small anode are collected.
A third approach employs a structure in which the anode of a planar CdZnTe detector is segmented into an array of very small individual detectors (pixels), with the cathode remaining as a single, continuous electrode. See H. Barrett, et al., "Charge Transport in Arrays of Semiconductor Gamma-Ray Detectors," Phys. Rev. Let. 75 (2). pp. 156-159 (1995). Here, each pixel is connected to an external signal conditioning circuit. Charge induced by an electron cloud is shared among the pixels and is very small on any specific pixel until the charge is very near the pixel.
This third approach also suffers from significant limitations. First, it is only useful for an array of very small pixels. Thus, this approach cannot be used for single-element detectors. Second, this approach is not applicable to detector arrays with pixel sizes of a millimeter or more, as used in nuclear medical imagers.
A fourth approach employs an anode patterned into an interleaved grid structure, with the cathode remaining planar. See, e.g., P. N. Luke, "Unipolar Charge Sensing with Co-Planar Electrodes--Application to Semiconductor Detectors," IEEE Tran. Nucl. Science, vol. 42, No. 4, at 207-213 (1 995). In the fLuke approach, one set of anode grids is maintained at a slightly higher voltage than the other. A train of signal conditioning electronics is connected to each set of grids, and the difference between the outputs from these trains constitutes the final output signal. With this arrangement, when the charge cloud is far from the grids, the difference-signal between the grid outputs is zero. As the cloud approaches the grids, the induced charge on one grid rises rapidly, while the charge induced in the other grid drops rapidly. The difference signal is then a measure of the full charge in the electron cloud, independent of the position of the ionizing event.
The Luke approach, however, also suffers from drawbacks. First, the grid structure is relatively complex and would be difficult, if not impossible, to use in detector arrays. Second, the grids require two separate amplifying chains, plus a difference amplifier, which add significantly to the complexity and cost of manufacture. This circuitry would also be very difficult to implement in multi-channel integrated circuits needed in detector array structures.
A final approach to reducing low-energy tailing may be implicit in the design of silicon drift chambers. See, e.g., E. Gatti & P. Rehak, "Semiconductor Drift Chamber--An Application of a Novel Charge Transport Scheme." Nucl. Inst. & Methods in Physics Research, vol. 225, at 608-614 (1984). A semiconductor drift chamber is based on the principle that a thin, large area semiconductor wafer, with rectifying junctions implanted on both surfaces, can be fully depleted through a small anode contact. The depletion field confines electrons generated by an ionizing particle in a buried potential channel parallel to the surface. An electrostatic field (drift field) parallel to the surface is independently superimposed and transports the electrons along the buried potential channel toward a collecting electrode. In addition, the capacitance of the collecting electrode is very low and independent of the active area of the detector. It has been suggested that drift chambers can be made from a variety of semiconductors. They have been implemented successfully with 300-micron-thick high-resistivity (10.sup.4 to 10.sup.5 ohm-cm) silicon wafers. Such drift chambers are used as high-resolution position-sensing detectors for particle physics.
The silicon drift chamber approach also suffers from several drawbacks. First, rectifying junction contacts must be used to generate the depletion field and the drift field. Because of the limitation of the breakdown voltage of these junction contacts, the magnitude of usable voltage is limited. This in turn limits the thickness of the wafer that can be used for the drift detector. Second, in order to transport charge effectively in the thin channel, a uniform drift field must be applied. A large number of junction contacts, each with a carefully controlled, fixed voltage, is required on the wafer to generate this uniform drift field. This adds significantly to the manufacturing cost and the complexity of using the detector. Third, because of the limitation of the thickness of the detector and the low Z (.about.14) of the semiconductor material used, the detection efficiency for x rays and gamma rays is very low for energies above 10-20 keV.
Both silicon detectors and detectors made of high resistivity materials, such as CdTe and CdZnTe, have employed "guard rings" around the signal electrode. A guard ring is normally kept at the same potential as the signal electrode and is used primarily to prevent dark current from the edge of a detector from reaching the signal electrode, thereby reducing the signal-to-noise ratio of the measurement. The guard ring does not significantly reduce low-energy tailing encountered in semiconductor detectors.
Therefore, a need exists for a semiconductor radiation detector that minimizes low-energy tailing and that obviates the disadvantages and drawbacks of conventional radiation detectors. The present invention provides such a radiation detector.